Efficacy of Epidural Blood Patching or Surgery in Spontaneous Intracranial Hypotension: A Systematic Review and Evidence Map

DISCUSSION

We identified and reviewed 139 studies that reported the efficacy of surgery and EBP, the 2 most common treatments for patients with SIH. We found no studies providing level 1 evidence for the efficacy of either treatment and identified no published clinical trials. Most published studies were retrospective, typically either case series or smaller cohort studies. In general, important study details were often missing, including methodologic details such as how a diagnosis of SIH was established and what specific subtypes of CSF leaks were being treated. Furthermore, procedural characteristics were often lacking, which made it difficult to determine important information such as whether imaging guidance or targeting was used in EBP or exactly what type of surgery was performed. Finally, patient-outcome assessment typically used subjective, nonvalidated measures collected at variable or unspecified time intervals.

The results of this evidence map suggest that the current efficacy estimates for EBP and surgery in SIH should be interpreted with caution. The incomplete description of patient characteristics in many of the included studies makes it difficult to understand how the study cohort compares with the broader SIH population, limiting applicability. For example, in some studies, the methods for establishing an SIH diagnosis were ambiguous, leading to uncertainty about the diagnosis itself. Additionally, we found that 77.7% of the studies did not adequately describe the subtypes of CSF leaks that were being treated. There may prove to be substantial differences in the treatment outcomes between patients with different subtypes of CSF leaks, which could skew the results of individual studies, possibly leading to spurious conclusions. Similarly, the absence of key procedural details in many instances makes it difficult, if not impossible, to understand the impact of certain decisions on the efficacy of treatments, such as whether to use a targeted approach or to include fibrin glue in EBP. Finally, the use of subjective, nonvalidated outcome measures without prespecified or uniform collection times leads to uncertainty about the usefulness of the collected data. Without establishing content validity, a chosen outcome measure may not truly capture the intended information of interest. For example, the use of outcome measures focused on patients’ pain may not accurately reflect the severity of disease if morbidity is due to cranial nerve dysfunction, fatigue, or cognitive dysfunction. Moreover, collecting outcomes at variable time points makes it challenging to compare cohorts of patients and could introduce bias (eg, outcomes only collected when the patient is symptomatic and returning to clinic).

To our knowledge, this represents the first systematic review to map the evidence for the efficacy of EBP and surgery in the treatment of SIH. We identified 3 previously published systematic reviews of SIH, but none of these mapped the current evidence to describe characteristics of the literature nor did they identify evidence clusters and knowledge gaps. Two also included meta-analyses.^19,20 The article by D’Antona et al¹⁹ is a general review of SIH containing a small subsection on treatment, which provides some summary estimates about conservative approaches, EBP, and surgery. The study by Signorelli et al²⁰ focused on factors affecting the outcome of EBP in SIH. Their review and analysis included only 6 articles, all retrospective. This small number was presumably due to the strict inclusion criteria used, such as the need for the studies to be comparative. This finding is in keeping with our findings that there are few prospective or comparative studies of EBP. Shlobin et al²¹ published a systematic review of patients with SIH with CSF-to-venous fistulas. The purpose of their review was to provide a comprehensive picture of CSF-to-venous fistulas, rather than to focus solely on treatment or map the evidence, as in the present study. They found 16 articles that met the inclusion criteria and performed a patient-level meta-analysis on 18 patients from 7 of these studies. Similarly, they found either only retrospective case series or cohort studies.

Given the considerable clinical and methodologic heterogeneity of the SIH literature, there was resultant statistical heterogeneity in many of the meta-analyses in these 2 studies, which raises some concern about the pooled estimates because they could lead to misleading conclusions. This heterogeneity is reflected in large I² values (a measure of variation due to study heterogeneity rather than chance) for the meta-analyses, many >75% (considered high heterogeneity) and some in excess of 95%.^19,20 In fact, very few published meta-analyses have I² > 90%.²² Sources of heterogeneity are not identified, calling into question the value of the pooled estimates. In addition, many of the meta-analyses consisted of predominantly small studies. Although random effects analysis is appropriate, it can lend undue weight to smaller studies, skewing the estimated summary effect. Given the resultant uncertainty regarding the reported estimates, caution should be used before changing practice on the basis of these results.

The current study adds to these prior systematic reviews by providing clear descriptions of study design and methodology, including details about SIH diagnostic criteria, patient characteristics, procedure specifics, and the outcome measures used. Furthermore, this investigation maps the breadth of the literature rather than focusing on a narrow question in a small subset of studies. It, therefore, provides a better perspective to identify knowledge gaps and to suggest designs that might address these unmet needs. The results of this investigation reveal several evidence gaps and clear opportunities for methodologic improvement, which, when addressed through future research, will serve to improve the state of the literature and advance our knowledge about SIH treatment efficacy. First, the dearth of prospective studies and the absence of any level 1 evidence or clinical trials demonstrate a clear unmet need. Similarly, there are few comparative studies. Thus, future research should be directed toward designing comparative prospective studies beginning with cohort designs followed by randomized controlled trials, preferably with a multicenter approach to maximize patient enrollment and generalizability. These efforts may be best facilitated through development of a research infrastructure among a consortium of dedicated research institutions. Furthermore, there are very few studies investigating surgical treatments of SIH. Dedicated effort to determine the efficacy of SIH treatments other than EBP (ie, surgery and the newly developed venous embolization of CSF-venous fistulas) should be pursued.²³ Second, future research should endeavor to reduce the potential for bias through adherence to sound methodologic practices and to improve interpretability through the clear reporting of key components of the study design. We recommend that objective SIH diagnostic criteria be used for patient inclusion, which should follow the current ICHD criteria at the time of the study (currently ICHD-3).¹⁶ Similarly, investigators should report the subtypes of spinal CSF leaks included in their cohort and whether there are efficacy differences among these subgroups of patients with SIH.

The following key procedural details should also be included at a minimum so that results may be better interpreted. For EBP, we strongly recommend including whether imaging guidance was used and the type (eg, CT fluoroscopy), whether the needles were targeted to the site of a known CSF leak or nontargeted, and the type of patch material used (eg, blood, fibrin glue, or both). For surgery, we recommend describing the operative approach or approaches in detail or providing references to previously described techniques. Critically, we recommend reporting symptom and disability outcome measures that are validated for use in SIH and that these measures be collected at prespecified uniform time points. To our knowledge, there are currently no validated outcome measures for use in SIH. Thus, effort should be directed toward validating existing measures or generating new ones. Until these efforts are completed, we recommend using objective measures validated for use in other headache disorders, such as the Headache Impact Test-6 and EuroQol-5D (https://euroqol.org/).^24,25 Finally, the use of changes in imaging (eg, resolution of SIH findings on brain MR imaging) as an outcome measure should include either an objective scoring system or description.¹⁸

This study is a protocol-driven, methodologically rigorous evidence map. However, there are several potential limitations to this investigation. It is possible that our search inadvertently excluded studies that would have otherwise met the criteria for inclusion. This issue could have occurred due to the use of narrower search terms that included the “spontaneous” concept. However, we followed best practices for search development and reviewed a random sample of studies from the more expansive search to ensure that the probability of this exclusion was minimal, at best. Furthermore, we did not search the gray literature because this was deemed unlikely to have identified relevant studies and, therefore, did not warrant the substantial additional effort. This investigation did not include the few studies that reported the efficacy of transvenous embolization of CSF-to-venous fistulas because this newer treatment method was developed after the protocol for this review was established and there are so few publications on this promising new intervention to date. Additionally, the authors recognize that some of our recommendations are dependent on presumptions that are yet to be proved. For example, if nontargeted patching is equivalent in efficacy to targeted patching, then identifying the CSF leak subtype and localizing the leak with advanced myelography techniques may not be necessary. Finally, given that evidence mapping is a relatively new approach to critically appraising the literature, the standards are evolving. We followed current best practices, but these may be subject to change.